REQUESTED TABLE DEFINITIONS
Table "public.call_record"
Column | Type | Modifiers
-----------------+------------------------+---------------
cntrct_id | character varying(15) | not null
call_regard | text |
port_type | character varying(9) |
inst | text |
info_taken | character varying(40) |
log_date | date | not null
log_time | time without time zone | not null
act_taken | text |
use_material | text |
targ_pest | integer |
work_comp_by | text |
emp_no | integer |
comp_date | date |
job_start_time | time without time zone |
job_leave_time | time without time zone |
comp_val | boolean | default false
fti_call_regd | public.tsvector |
fti_inst | public.tsvector |
fti_act_take | public.tsvector |
route | character(3) |
act_port | text |
targ_pest_opt | text |
call_regard_opt | text |
targpest_other | text |
date_sched | date |
custord_num | integer |
dist_id | integer |
phone_slot | integer | default 0
Indexes:
"call_record_pkey" PRIMARY KEY, btree (cntrct_id, log_date, log_time)
"route_index" hash (route)
Check constraints:
"call_record_targ_pest_check" CHECK (targ_pest <= 100)
"call_record_targ_pest_check1" CHECK (targ_pest >= 0)
Table "public.per_call"
Column | Type | Modifiers
---------+----------------------+-----------
dist_id | character varying(2) |
route | character varying(2) |
type | character(1) |
total | integer |
I need to get data from 2 tables and print it in a single report. The report should look like this:
district | route | type | total | callbacks
| 01 | T | 12 | 5
| 02 | P | 0 | 0
| 03 | P | 3 | 1
2 | 01 | T | 4 | 1
| 02 | T | 1 | 0
| 03 | P | 0 | 0
etc... (this is theoretical sample data)
So, in essence I need to get the dist_id, route, type, and count(*) from the table per_call and the count of call_backs from the table call_record
PROBLEM: looping through tables makes it go glacially slow. How can I adjust the following PSQL query so that I don't have to loop and I can echo the tabular data properly?
Let me know if anything is opaque and I will try to clarify
echo '<table align="center" border = 2>
<th>DISTRICT</th>
<th>ROUTE</th>
<th>TYPE</th>
<th>TOTAL</th>
<th>CALL BACKS</th>';
$SQL = " SELECT per_call.dist_id, per_call.route, per_call.type, per_call.total
FROM per_call, call_record
WHERE TRUE ";
if($type == 'termite'){
$SQL = $SQL." AND per_call.type = 'T' ";
}
else{
$SQL = $SQL." AND per_call.type = 'P' ";
}
$SQL = $SQL." AND call_record.dist_id = per_call.dist_id
AND call_record.log_date >= '$startDate'
AND call_record.log_date <= '$endDate'
ORDER BY per_call.dist_id, per_call.route, per_call.type ASC ";
echo $SQL;
/*AND call_record.log_date = '$startDate'
AND call_record.log_date = '$endDate'*/
$Q = pg_query($connect,$SQL);
while($row = pg_fetch_row($Q)){
$dist = $row[0];
$route = $row[1];
$type = $row[2];
$total = $row[3];
echo '<tr>';
echo '<td align="center">'.$dist.'</td>';
echo '<td align="center">'.$route.'</td>';
echo '<td align="center">'.$type.'</td>';
echo '<td align="center">'.$total.'</td>';
$SQL2 = "SELECT COUNT(*)
FROM call_record
WHERE dist_id = $dist
AND route = '$route'
AND substring(cntrct_id from 2 for 1) = '$type'
AND substring(call_regard_opt from 2 for 1) = '1'
";
$Q2 = pg_query($connect,$SQL2);
$row2 = pg_fetch_row($Q2);
$callbacks = $row2[0];
echo '<td align="center">'.$callbacks.'</td>';
echo '</tr>';
}
echo "</table>";
select pc.dist_id, pc.route, pc.type, pc.total,
count(
substring(cntrct_id from 2 for 1) = '$type'
AND substring(call_regard_opt from 2 for 1) = '1'
or null
) callbacks
from
per_call pc
inner join
call_record cr on cr.dist_id = pc.dist_id
where cr.log_date between '$startdate' and cr.log_date <= '$enddate'
group by pc.dist_id, pc.route, pc.type, pc.total
order by pc.dist_id, pc.route, pc.type asc
Unless index problems, your query is not badly written, so it can't be really optimized from a query form point of view. You can do that though which is cleaner :
$SQL = "SELECT per_call.dist_id, per_call.route, per_call.type, per_call.total
FROM per_call, call_record
WHERE call_record.dist_id = per_call.dist_id
AND per_call.type = '".($type == 'termite' ? "T" : "P")."'
AND call_record.log_date >= '$startDate'
AND call_record.log_date <= '$endDate'
ORDER BY per_call.dist_id, per_call.route, per_call.type"
Anyway, this is still vulnerable to SQL injections. Try using parameterized queries.
You don't have any indexes on dist_id in either table, which is going to make your join very slow. Add indexes on dist_id and see how much it improves.
Also, that query inside the loop is going to be the death of you, because you're going to be doing many many many queries. Work the inner query into your main query so you only execute one query in the database.